275 lines
12 KiB
Python
275 lines
12 KiB
Python
"""
|
|
TEE OFF - VEIEN TIL GOLF (VTG) SKRAPER MED GEMINI AI
|
|
---------------------------------------------------------------------------
|
|
Henter pris, beskrivelse (inkl. lånekøller/medlemskap) og kursdatoer fra VTG-sider.
|
|
Støtter kommaseparerte URL-er.
|
|
---------------------------------------------------------------------------
|
|
"""
|
|
|
|
import asyncio
|
|
import asyncpg
|
|
import os
|
|
import json
|
|
import argparse
|
|
from bs4 import BeautifulSoup
|
|
from playwright.async_api import async_playwright
|
|
import google.generativeai as genai
|
|
from dotenv import load_dotenv
|
|
from env_config import get_database_url
|
|
from scrape_utils import ProgressCallback, emit_progress, make_progress_event, parse_llm_json
|
|
|
|
load_dotenv()
|
|
|
|
DB_URL = get_database_url()
|
|
GEMINI_API_KEY = os.getenv("GEMINI_API_KEY")
|
|
|
|
if not GEMINI_API_KEY:
|
|
raise ValueError("🚨 GEMINI_API_KEY mangler i .env filen!")
|
|
|
|
genai.configure(api_key=GEMINI_API_KEY)
|
|
model = genai.GenerativeModel('gemini-2.5-flash')
|
|
|
|
async def fetch_page_text(url: str, browser) -> str:
|
|
url = url.strip()
|
|
if not url.startswith("http"):
|
|
return ""
|
|
|
|
print(f" 🌐 Laster inn: {url}")
|
|
try:
|
|
page = await browser.new_page()
|
|
await page.goto(url, wait_until="domcontentloaded", timeout=15000)
|
|
html_content = await page.content()
|
|
await page.close()
|
|
|
|
soup = BeautifulSoup(html_content, 'html.parser')
|
|
for script in soup(["script", "style", "nav", "footer", "header"]):
|
|
script.extract()
|
|
|
|
return soup.get_text(separator=' ', strip=True)
|
|
except Exception as e:
|
|
print(f" ❌ Feil ved lasting av {url}: {e}")
|
|
return ""
|
|
|
|
def analyze_vtg_with_gemini(text: str, club_name: str) -> dict | None:
|
|
print(f" 🧠 Sender {len(text)} tegn til Gemini for VTG-analyse...")
|
|
|
|
prompt = f"""
|
|
Du er en ekspert på norske golfklubber. Din oppgave er å lese en lang tekst fra nettsidene til "{club_name}" og koke dette ned til essensen om deres "Veien til Golf" (VTG) nybegynnerkurs.
|
|
|
|
OPPGAVER:
|
|
1. Finn standardprisen for VTG-kurset for en vanlig voksen person. (Returner KUN tallet).
|
|
2. Skriv en KOMPRIMERT, selgende beskrivelse (maks 3-4 setninger). Du MÅ inkludere informasjon om:
|
|
- Er lån av køller/utstyr inkludert i kurset?
|
|
- Inkluderer prisen et medlemskap/spillerett i klubben (og ev. for hvor lenge)?
|
|
- Hva er omfanget? (F.eks. "12 timer praksis pluss e-læring").
|
|
Ignorer uvesentlig støy og lange historiske utgreiinger.
|
|
3. Finn alle kommende kursdatoer. Finn startdato/sluttdato for hvert kurs, og noter status ("Ledig", "Fulltegnet", "Venteliste").
|
|
|
|
TEKST FRA NETTSIDEN:
|
|
{text}
|
|
|
|
OPPGAVE:
|
|
Returner KUN et gyldig JSON-objekt med nøyaktig følgende struktur:
|
|
{{
|
|
"foreslatt_vtg_pris": 1990,
|
|
"foreslatt_vtg_beskrivelse": "Kurset går over 12 timer inkludert obligatorisk e-læring. Lån av golfkøller er inkludert under hele kurset, og prisen gir deg også fritt spill og medlemskap ut året.",
|
|
"foreslatt_vtg_datoer": [
|
|
{{"dato": "12.-14. mai", "status": "Fulltegnet"}},
|
|
{{"dato": "5.-7. juni", "status": "Ledig"}}
|
|
],
|
|
"ai_begrunnelse": "Fant voksenpris på 1990,-. Teksten nevnte eksplisitt at medlemskap ut året er med i prisen, og at man får låne utstyr."
|
|
}}
|
|
Merk: Sett foreslatt_vtg_pris til null (null) hvis du ikke finner den. Hvis du ikke finner datoer, la listen være tom [].
|
|
"""
|
|
|
|
try:
|
|
response = model.generate_content(prompt)
|
|
parsed = parse_llm_json(response.text)
|
|
return parsed if isinstance(parsed, dict) else None
|
|
except Exception as e:
|
|
print(f" ❌ AI-analyse feilet: {e}")
|
|
return None
|
|
|
|
async def run_vtg_scraper(facility_ids=None, progress_callback: ProgressCallback | None = None):
|
|
print("🚀 Starter Veien til Golf (VTG) skraperen...")
|
|
conn = await asyncpg.connect(DB_URL)
|
|
facilities = []
|
|
analyzed_count = 0
|
|
saved_count = 0
|
|
skipped_count = 0
|
|
failed_count = 0
|
|
|
|
try:
|
|
query = "SELECT id, name, vtg_lenke FROM facilities WHERE vtg_lenke IS NOT NULL AND vtg_lenke != ''"
|
|
if facility_ids:
|
|
query += f" AND id IN ({','.join(map(str, facility_ids))})"
|
|
|
|
facilities = await conn.fetch(query)
|
|
total_facilities = len(facilities)
|
|
print(f"📋 Fant {total_facilities} anlegg å skrape.")
|
|
await emit_progress(
|
|
progress_callback,
|
|
progress_total=total_facilities,
|
|
progress_completed=0,
|
|
progress_ok=0,
|
|
progress_failed=0,
|
|
progress_skipped=0,
|
|
event=make_progress_event(
|
|
facility_id=None,
|
|
facility_name="VTG",
|
|
outcome="info",
|
|
message=f"Starter VTG-skraping for {total_facilities} anlegg.",
|
|
processed=0,
|
|
total=total_facilities,
|
|
),
|
|
)
|
|
|
|
async with async_playwright() as p:
|
|
browser = await p.chromium.launch(headless=True)
|
|
|
|
for index, facility in enumerate(facilities, start=1):
|
|
fac_id = facility['id']
|
|
name = facility['name']
|
|
urls_raw = facility['vtg_lenke']
|
|
|
|
print(f"\n▶️ Behandler VTG for: {name} (ID: {fac_id})")
|
|
await emit_progress(
|
|
progress_callback,
|
|
current_facility_id=fac_id,
|
|
current_facility_name=name,
|
|
event=make_progress_event(
|
|
facility_id=fac_id,
|
|
facility_name=name,
|
|
outcome="info",
|
|
message="Starter henting av VTG-sider.",
|
|
processed=index - 1,
|
|
total=total_facilities,
|
|
),
|
|
)
|
|
|
|
urls = [u.strip() for u in urls_raw.split(',')]
|
|
combined_text = ""
|
|
try:
|
|
for idx, url in enumerate(urls, 1):
|
|
page_text = await fetch_page_text(url, browser)
|
|
if page_text:
|
|
combined_text += f"\n\n--- TEKST FRA SIDE {idx} ({url}) ---\n{page_text}"
|
|
|
|
if len(combined_text) < 50:
|
|
print(" ⚠️ Fant for lite tekst, hopper over.")
|
|
skipped_count += 1
|
|
await emit_progress(
|
|
progress_callback,
|
|
progress_completed=index,
|
|
progress_ok=saved_count,
|
|
progress_failed=failed_count,
|
|
progress_skipped=skipped_count,
|
|
current_facility_id=fac_id,
|
|
current_facility_name=name,
|
|
event=make_progress_event(
|
|
facility_id=fac_id,
|
|
facility_name=name,
|
|
outcome="warning",
|
|
message="Hoppet over fordi det ble funnet for lite tekst på VTG-sidene.",
|
|
processed=index,
|
|
total=total_facilities,
|
|
),
|
|
)
|
|
continue
|
|
|
|
draft_data = analyze_vtg_with_gemini(combined_text[:25000], name)
|
|
if not draft_data:
|
|
failed_count += 1
|
|
await emit_progress(
|
|
progress_callback,
|
|
progress_completed=index,
|
|
progress_ok=saved_count,
|
|
progress_failed=failed_count,
|
|
progress_skipped=skipped_count,
|
|
current_facility_id=fac_id,
|
|
current_facility_name=name,
|
|
event=make_progress_event(
|
|
facility_id=fac_id,
|
|
facility_name=name,
|
|
outcome="error",
|
|
message="AI-analysen ga ikke et gyldig VTG-utkast.",
|
|
processed=index,
|
|
total=total_facilities,
|
|
),
|
|
)
|
|
continue
|
|
|
|
analyzed_count += 1
|
|
found_dates = len(draft_data.get('foreslatt_vtg_datoer', []))
|
|
print(f" ✅ AI fant pris: {draft_data.get('foreslatt_vtg_pris')}, og {found_dates} datoer.")
|
|
|
|
await conn.execute("""
|
|
UPDATE facilities
|
|
SET vtg_draft = $1::jsonb
|
|
WHERE id = $2
|
|
""", json.dumps(draft_data), fac_id)
|
|
|
|
print(" 💾 VTG-utkast lagret i databasen!")
|
|
saved_count += 1
|
|
await emit_progress(
|
|
progress_callback,
|
|
progress_completed=index,
|
|
progress_ok=saved_count,
|
|
progress_failed=failed_count,
|
|
progress_skipped=skipped_count,
|
|
current_facility_id=fac_id,
|
|
current_facility_name=name,
|
|
event=make_progress_event(
|
|
facility_id=fac_id,
|
|
facility_name=name,
|
|
outcome="success",
|
|
message=f"Utkast lagret med pris {draft_data.get('foreslatt_vtg_pris') or 'ukjent'} og {found_dates} kursdatoer.",
|
|
processed=index,
|
|
total=total_facilities,
|
|
),
|
|
)
|
|
except Exception as e:
|
|
failed_count += 1
|
|
print(f" ❌ Uventet feil for {name}: {e}")
|
|
await emit_progress(
|
|
progress_callback,
|
|
progress_completed=index,
|
|
progress_ok=saved_count,
|
|
progress_failed=failed_count,
|
|
progress_skipped=skipped_count,
|
|
current_facility_id=fac_id,
|
|
current_facility_name=name,
|
|
event=make_progress_event(
|
|
facility_id=fac_id,
|
|
facility_name=name,
|
|
outcome="error",
|
|
message=f"Feilet under behandling: {str(e).splitlines()[0]}",
|
|
processed=index,
|
|
total=total_facilities,
|
|
),
|
|
)
|
|
|
|
await browser.close()
|
|
|
|
finally:
|
|
await conn.close()
|
|
print("\n🏁 Skraping fullført.")
|
|
|
|
return {
|
|
"processed_facilities": len(facilities),
|
|
"analyzed_facilities": analyzed_count,
|
|
"saved_drafts": saved_count,
|
|
"skipped_facilities": skipped_count,
|
|
"failed_facilities": failed_count,
|
|
}
|
|
|
|
if __name__ == "__main__":
|
|
parser = argparse.ArgumentParser(description="Skrap VTG via AI.")
|
|
parser.add_argument("--ids", type=str, help="Kommaseparert liste med facility IDs (eks: 1,5,12)")
|
|
args = parser.parse_args()
|
|
|
|
ids_to_scrape = None
|
|
if args.ids:
|
|
ids_to_scrape = [int(x.strip()) for x in args.ids.split(",")]
|
|
|
|
asyncio.run(run_vtg_scraper(ids_to_scrape))
|